When Apple unveiled its innovative artificial intelligence (Apple Intelligence) two years ago, it made very bold promises. The most prominent promise was to deliver a smarter and more personalized version of “Siri.” Despite the obstacles and delays that faced this project, Apple confirmed at the time that the AI features would rely on a unique cloud infrastructure built on a single core pillar: maximum user privacy, known as “Private Cloud Compute.”
When Apple Silicon cloud power is not enough!
Apple’s unannounced direction was to run its Private Cloud Compute servers using its own super-processors, such as the M2 Ultra. However, it seems that technical reality imposed its difficult conditions; despite the amazing efficiency of Apple Silicon processors in running iPhones, iPads, and Macs, and processing AI locally on the device, they were not fully prepared to keep up with the massive requirements needed by cloud servers to run huge and complex AI models compared to solutions available on the market.
After difficult periods and delays in delivering the new “Siri,” Apple was forced to make decisive decisions that included restructuring some of its leadership and acknowledging that it needed external help from industry giants. This decision led to fierce competition between the three biggest players in the AI market (Anthropic, OpenAI, and Google), with the competition ending in the selection of Google’s Gemini model to be the cornerstone of the new Siri project.
Nvidia Blackwell B200 processors at the heart of Siri’s cloud brain
Here comes the most prominent turning point; according to a new and reliable report from The Information, sources confirmed that Apple’s new Private Cloud Compute (PCC) architecture will not rely on Apple Silicon as planned, but will instead rely on Nvidia’s super-powerful Blackwell B200 processors! These are the same processors that Google relies on to run its most powerful servers for processing intelligent Gemini models.
The report indicated that some user requests directed to the smart assistant Siri will be processed within Google Cloud via licensed versions of the Gemini model. But how will Apple keep its strict promises regarding the privacy of sensitive user data while using another company’s cloud?
Privacy first: How does Apple ensure our data security?
Of course, Apple would not have taken such a step without strict guarantees. The report explained that the company has tested and adopted a very advanced protection technology from Nvidia known as “Confidential Computing.” This unique technology works by fully encrypting data, requests, and AI models at the hardware level during processing, ensuring that none of the user queries are saved or recorded, and that even Google or Nvidia cannot view them.
Therefore, Apple will continue to use its “Private Cloud Compute” brand without change, as security remains guaranteed even if the processing power changes to become the exceptional strength of Nvidia instead of Apple’s own chips, proving once again that the race to the top in the world of AI sometimes requires unexpected alliances and quick tactical concessions.
Source:
Leave a Reply